Hang Thanh Bui
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Hang Thanh Bui has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- Arash Mahboubi
UNSW Canberra
8 shared publications - Hamed Aboutorab
UNSW Canberra
8 shared publications - Seyit Camtepe
CSIRO Data61
7 shared publications - Keyvan Ansari
Murdoch University
5 shared publications - Khanh Luong
Queensland University of Technology
5 shared publications - Michael Bewong
School of Computing, Mathematics and Engineering, Charles Sturt University, Wagga Wagga, NSW 2678, Australia
3 shared publications
Latest Publications
- Open Access
- 11 November 2026
- Published by Springer Nature in Research Square
CarbonCrediBERT: Expert-Guided Few-Shot and Selective Classification of Carbon Credit Disclosures
Corporate reports increasingly discuss carbon credits in relation to net-zero targets, offsetting claims, and climate transition plans, particularly as climate-related financial disclosure requirements expand. Analysing these disclosures at scale remains difficult because terms such as credit, offset, and net also occur in unrelated financial, accounting, and operational contexts. Expert-labelled data are also costly to produce and often contain substantial class imbalance. We propose CarbonCrediBERT, an expert-guided natural language processing framework that classifies paragraphs according to carbon credit relevance, sentiment, and reported credit use, while referring uncertain cases for human review. CarbonCrediBERT combines an expert defined taxonomy, hard-negative examples drawn from closely related but nonrelevant passages, and classification methods suited to small expert-labelled datasets. Using annual, climate, sustainability, and ESG reports published by ASX100 companies in 2024–2025, the framework produced 8,299 keyword-matched sentence records and a balanced candidate pool of 15,026 paragraphs. We evaluated ClimateBERT and DistilRoBERTa fine-tuning and SetFit–MiniLM few-shot learning using three task-specific datasets of 100 expert-labelled paragraphs each. Five-fold development cross-validation selected fine-tuned ClimateBERT for carbon credit relevance, achieving 93.8% accuracy, and SetFit–MiniLM for sentiment and reported credit use. These results show that broad climate-language detection alone is insufficient for identifying carbon credit disclosures and that no single modelling approach performs best across all three tasks. They also highlight the continuing importance of expert judgement, particularly for sentiment and ambiguous statements about credit use. Future work will expand the domain corpus and expert-labelled evidence and evaluate transfer across companies, reporting years, and jurisdictions, establishing a pathway towards a reusable CarbonCrediBERT foundation model for carbon credit disclosure analysis.
- 11 November 2026
- Published by Springer Nature in Research Square
Open AccessCarbonCrediBERT: Expert-Guided Few-Shot and Selective Classification of Carbon Credit Disclosures
Corporate reports increasingly discuss carbon credits in relation to net-zero targets, offsetting claims, and climate transition plans, particularly as climate-related financial disclosure requirements expand. Analysing these disclosures at scale remains difficult because terms such as credit, offset, and net also occur in unrelated financial, accounting, and operational contexts. Expert-labelled data are also costly to produce and often contain substantial class imbalance. We propose CarbonCrediBERT, an expert-guided natural language processing framework that classifies paragraphs according to carbon credit relevance, sentiment, and reported credit use, while referring uncertain cases for human review. CarbonCrediBERT combines an expert defined taxonomy, hard-negative examples drawn from closely related but nonrelevant passages, and classification methods suited to small expert-labelled datasets. Using annual, climate, sustainability, and ESG reports published by ASX100 companies in 2024–2025, the framework produced 8,299 keyword-matched sentence records and a balanced candidate pool of 15,026 paragraphs. We evaluated ClimateBERT and DistilRoBERTa fine-tuning and SetFit–MiniLM few-shot learning using three task-specific datasets of 100 expert-labelled paragraphs each. Five-fold development cross-validation selected fine-tuned ClimateBERT for carbon credit relevance, achieving 93.8% accuracy, and SetFit–MiniLM for sentiment and reported credit use. These results show that broad climate-language detection alone is insufficient for identifying carbon credit disclosures and that no single modelling approach performs best across all three tasks. They also highlight the continuing importance of expert judgement, particularly for sentiment and ambiguous statements about credit use. Future work will expand the domain corpus and expert-labelled evidence and evaluate transfer across companies, reporting years, and jurisdictions, establishing a pathway towards a reusable CarbonCrediBERT foundation model for carbon credit disclosure analysis.
- Open Access
- 10 October 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Privacy and Security
- Vol. 29 (4) , 1-36
Privacy Risks and Leakage Pathways in Agentic and Multi-Agent AI Systems
Agentic and multi-agent AI systems are increasingly deployed as part of operational systems, introducing autonomous planning, delegation, persistent memory, tool use, and inter-agent coordination into data-driven decision making. These architectural properties challenge prevailing privacy assumptions because leakage may arise structurally from autonomy, coordination, persistence, and composition rather than only from isolated model outputs. This article provides a structured synthesis of privacy risks and leakage pathways for agentic and multi-agent AI systems. We introduce a unifying abstraction that combines decentralized decision making with trace-based information-flow semantics, enabling reasoning about autonomy, coordination, persistent state, observability, and adversarial trace projections. Using this abstraction, we organize recurring privacy risks into five leakage pathways: autonomy-induced overexposure, coordination-driven information propagation, memory and persistent-state leakage, emergent and compositional inference, and observability and audit leakage. We further analyze why established privacy-enhancing technologies provide incomplete coverage in agentic settings, where static computation boundaries, single-agent assumptions, ephemeral state, and non-emergent behavior often fail. To ground the analysis, we implement a controlled multi-agent privacy sandbox with privacy-aware agents, tool-mediated workflows, adversarial trace projections, and LLM-backed validation. The results show that policy-checked privacy-aware execution reduces direct disclosure and adversarial inference advantage, while residual tool-mediated and semantic-hint leakage remains. The article concludes by proposing new privacy requirements for agentic and multi-agent AI systems.
- 10 October 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Privacy and Security
- Vol. 29 (4) , 1-36
Open AccessPrivacy Risks and Leakage Pathways in Agentic and Multi-Agent AI Systems
Agentic and multi-agent AI systems are increasingly deployed as part of operational systems, introducing autonomous planning, delegation, persistent memory, tool use, and inter-agent coordination into data-driven decision making. These architectural properties challenge prevailing privacy assumptions because leakage may arise structurally from autonomy, coordination, persistence, and composition rather than only from isolated model outputs. This article provides a structured synthesis of privacy risks and leakage pathways for agentic and multi-agent AI systems. We introduce a unifying abstraction that combines decentralized decision making with trace-based information-flow semantics, enabling reasoning about autonomy, coordination, persistent state, observability, and adversarial trace projections. Using this abstraction, we organize recurring privacy risks into five leakage pathways: autonomy-induced overexposure, coordination-driven information propagation, memory and persistent-state leakage, emergent and compositional inference, and observability and audit leakage. We further analyze why established privacy-enhancing technologies provide incomplete coverage in agentic settings, where static computation boundaries, single-agent assumptions, ephemeral state, and non-emergent behavior often fail. To ground the analysis, we implement a controlled multi-agent privacy sandbox with privacy-aware agents, tool-mediated workflows, adversarial trace projections, and LLM-backed validation. The results show that policy-checked privacy-aware execution reduces direct disclosure and adversarial inference advantage, while residual tool-mediated and semantic-hint leakage remains. The article concludes by proposing new privacy requirements for agentic and multi-agent AI systems.
- Open Access
- 4 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 1074-1090
Setup Once, Secure Always: A Single-Setup Secure Federated Learning Aggregation Protocol with Forward and Backward Secrecy for Dynamic Users
- 4 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 1074-1090
Open AccessSetup Once, Secure Always: A Single-Setup Secure Federated Learning Aggregation Protocol with Forward and Backward Secrecy for Dynamic Users
- Open Access
- 13 August 2025
- Published by Elsevier in Internet of Things
- Vol. 33
The evolving threat landscape of botnets: Comprehensive analysis of detection techniques in the age of artificial intelligence
- 13 August 2025
- Published by Elsevier in Internet of Things
- Vol. 33
Open AccessThe evolving threat landscape of botnets: Comprehensive analysis of detection techniques in the age of artificial intelligence
- Open Access
- 15 July 2025
- Published by Elsevier in Journal of Information Security and Applications
- Vol. 93
ConceptUML: Multiphase unsupervised threat detection via latent concept learning, Hidden Markov Models and topic modelling
- 15 July 2025
- Published by Elsevier in Journal of Information Security and Applications
- Vol. 93
Open AccessConceptUML: Multiphase unsupervised threat detection via latent concept learning, Hidden Markov Models and topic modelling
- 7 July 2025
- Published by Springer Nature
- p. 399-414
Ransomware Encryption Detection: Adaptive File System Analysis Against Evasive Encryption Tactics
- 7 July 2025
- Published by Springer Nature
- p. 399-414
Ransomware Encryption Detection: Adaptive File System Analysis Against Evasive Encryption Tactics
- Open Access
Data Encryption Battlefield: A Deep Dive into the Dynamic Confrontations in Ransomware Attacks
In the rapidly evolving landscape of cybersecurity threats, ransomware represents a significant challenge. Attackers increasingly employ sophisticated encryption methods, such as entropy reduction through Base64 encoding, and partial or intermittent encryption to evade traditional detection methods. This study explores the dynamic battle between adversaries who continuously refine encryption strategies and defenders developing advanced countermeasures to protect vulnerable data. We investigate the application of online incremental machine learning algorithms designed to predict file encryption activities despite adversaries evolving obfuscation techniques. Our analysis utilizes an extensive dataset of 32.6 GB, comprising 11,928 files across multiple formats, including Microsoft Word documents (doc), PowerPoint presentations (ppt), Excel spreadsheets (xlsx), image formats (jpg, jpeg, png, tif, gif), PDFs (pdf), audio (mp3), and video (mp4) files. These files were encrypted by 75 distinct ransomware families, facilitating a robust empirical evaluation of machine learning classifiers effectiveness against diverse encryption tactics. Results highlight the Hoeffding Tree algorithms superior incremental learning capability, particularly effective in detecting traditional and AES-Base64 encryption methods employed to lower entropy. Conversely, the Random Forest classifier with warm-start functionality excels at identifying intermittent encryption methods, demonstrating the necessity of tailored machine learning solutions to counter sophisticated ransomware strategies.
Open AccessData Encryption Battlefield: A Deep Dive into the Dynamic Confrontations in Ransomware Attacks
In the rapidly evolving landscape of cybersecurity threats, ransomware represents a significant challenge. Attackers increasingly employ sophisticated encryption methods, such as entropy reduction through Base64 encoding, and partial or intermittent encryption to evade traditional detection methods. This study explores the dynamic battle between adversaries who continuously refine encryption strategies and defenders developing advanced countermeasures to protect vulnerable data. We investigate the application of online incremental machine learning algorithms designed to predict file encryption activities despite adversaries evolving obfuscation techniques. Our analysis utilizes an extensive dataset of 32.6 GB, comprising 11,928 files across multiple formats, including Microsoft Word documents (doc), PowerPoint presentations (ppt), Excel spreadsheets (xlsx), image formats (jpg, jpeg, png, tif, gif), PDFs (pdf), audio (mp3), and video (mp4) files. These files were encrypted by 75 distinct ransomware families, facilitating a robust empirical evaluation of machine learning classifiers effectiveness against diverse encryption tactics. Results highlight the Hoeffding Tree algorithms superior incremental learning capability, particularly effective in detecting traditional and AES-Base64 encryption methods employed to lower entropy. Conversely, the Random Forest classifier with warm-start functionality excels at identifying intermittent encryption methods, demonstrating the necessity of tailored machine learning solutions to counter sophisticated ransomware strategies.
- Open Access
A Survey on Unauthorized UAV Threats to Smart Farming
The integration of Internet of Things (IoT) and unmanned aerial vehicles (UAVs) in smart farming has revolutionized agricultural practices by enhancing monitoring, automation, and decision-making to improve agricultural productivity and sustainability. However, the widespread use of these technologies has also introduced new security challenges, particularly the risk of interference from unauthorized UAVs. This survey provides an analysis of the threats posed by unauthorized UAVs to smart farms, highlighting potential vulnerabilities such as data interception, communication jamming, and physical damage. This paper first explores recent advancements in IoT and UAV technologies, which are integral to the functioning of smart farms. Then, we present an analysis of unauthorized UAV threats to smart farms and evaluate the current state-of-the-art UAV countermeasure technologies. By examining these emerging threats and potential solutions, this survey aims to inform researchers, engineers, policymakers, and practitioners involved in smart farming about the critical need for enhanced anti-UAV systems. Additionally, it highlights the necessity for airspace management authorities to recognize the risks posed by unauthorized UAVs, invest resources in protective measures, and address the challenges associated with securing smart farms against unauthorized UAV threats.
Open AccessA Survey on Unauthorized UAV Threats to Smart Farming
The integration of Internet of Things (IoT) and unmanned aerial vehicles (UAVs) in smart farming has revolutionized agricultural practices by enhancing monitoring, automation, and decision-making to improve agricultural productivity and sustainability. However, the widespread use of these technologies has also introduced new security challenges, particularly the risk of interference from unauthorized UAVs. This survey provides an analysis of the threats posed by unauthorized UAVs to smart farms, highlighting potential vulnerabilities such as data interception, communication jamming, and physical damage. This paper first explores recent advancements in IoT and UAV technologies, which are integral to the functioning of smart farms. Then, we present an analysis of unauthorized UAV threats to smart farms and evaluate the current state-of-the-art UAV countermeasure technologies. By examining these emerging threats and potential solutions, this survey aims to inform researchers, engineers, policymakers, and practitioners involved in smart farming about the critical need for enhanced anti-UAV systems. Additionally, it highlights the necessity for airspace management authorities to recognize the risks posed by unauthorized UAVs, invest resources in protective measures, and address the challenges associated with securing smart farms against unauthorized UAV threats.
- 27 November 2024
- Published by Springer Nature
- p. 204-215
A Lightweight Detection of Sequential Patterns in File System Events During Ransomware Attacks
- 27 November 2024
- Published by Springer Nature
- p. 204-215
A Lightweight Detection of Sequential Patterns in File System Events During Ransomware Attacks
- Top Cited PapersOpen Access
- 12 February 2024
- Published by Elsevier in Computers & Security
- Vol. 140
Agriculture 4.0 and beyond: Evaluating cyber threat intelligence sources and techniques in smart farming ecosystems
- 12 February 2024
- Published by Elsevier in Computers & Security
- Vol. 140
Top Cited PapersOpen AccessAgriculture 4.0 and beyond: Evaluating cyber threat intelligence sources and techniques in smart farming ecosystems